Submit

Back to the channels

#primate

  1. * phaleth joined #primate
  2. phaleth archlinux updated to openssl 4 and things are breaking
  3. blue hi phaleth, wow
  4. phaleth hi blue
  5. phaleth on desktop I mean
  6. blue ya
  7. phaleth but I'm not gonna update the mail server yet
  8. blue ya
  9. blue [blue@vmd170320 ~]$ pacman -Q openssl
  10. blue openssl 4.1.0-1
  11. blue phaleth: does that mean we shouldn't reboot?
  12. blue I wasn't planning to, but need to know
  13. phaleth containers aren't affected by whatever is on host
  14. phaleth the only what is affected are sshd and podman on the host I think
  15. phaleth rest of everything that'd need openssl is either local only or containerized as far as I remember
  16. phaleth anyway, feel free to reboot
  17. phaleth archlinux updated to linux 7.3
  18. phaleth $ sudo podman exec 000001-repopack-site-haproxy /usr/bin/openssl version
  19. phaleth OpenSSL 3.5.9 29 Sep 2026 (Library: OpenSSL 3.5.9 29 Sep 2026)
  20. phaleth actually haproxy uses AWS-LC 4.0.0, which makes it openssl agnostic
  21. blue great, I'll boot later
  22. nevet [Flog releases] Flog 0.1.18 - https://repopack.com/api/v0/projects/flog/releases/products/flog/versions/0.1.18
  23. dreamreal blue: the rss for those releases should be a link to actual points in the changelog
  24. blue dreamreal: how do you mean?
  25. dreamreal like, the flog rss *link* is to the actual release data
  26. dreamreal follow that link, you get the download information, not anything about the release, no link to the changelog, nothing
  27. blue I have no idea what you mean, but this is too early in the day, clanker said
  28. blue "You could reply: “Got it—the RSS item should open the version-specific changelog section, not the release API response.”"
  29. blue did clanker get it?
  30. * blue needs a coffee
  31. blue wait, dreamreal
  32. blue why aren't you opening a task on flog project?
  33. dreamreal Because it's yet another task! grrr, fine
  34. dreamreal grrr, how do I create a task from the rp website
  35. blue you can't
  36. blue rp is your friend, dreamreal
  37. dreamreal That sounds like a repopack issue, too :D
  38. blue that's intentional though
  39. blue the gui is intentionally limited
  40. dreamreal Understood, although *I* think you shouldn't have that limit, because it means I get to login twice
  41. blue `rp task create --project flog --title <title> [--text <text> | --file <path>]
  42. blue `
  43. dreamreal I guess if you're pushing people to agent work, it's okay, and I also ended up filing a note because I expected an $EDITOR invocation with no issue text
  44. blue well I'm gonna fix the short login timeout. but ideally rp should always be logged in. I believe the timeout is currently 30 days, the token
  45. blue now THAT's an issue worth submitting, dreamreal
  46. dreamreal Every time I have any rp interaction I get to log in
  47. blue well it's currently at least one action every 24h, dreamreal, extensible up to 30 days
  48. blue I can up those limits
  49. blue but again: create tasks, I will fix. if I don't know of something that bothers you, I can't fix
  50. dreamreal what project name would this refer to? repopack?
  51. blue yes
  52. blue thanks dreamreal
  53. dreamreal I also wonder about how you'd resolve the "what project does this apply to" problem - rp project list gives me MY projects, but I don't know how I'd look up "repopack" or "flog" without knowing those are the project names
  54. blue this is a general discoverability problem. you'd have it with gh as well
  55. blue on repopack it's a bit easier because there are no orgs
  56. dreamreal Yep, agreed. That's why I wonder about it.
  57. blue generally speaking, since repopack is the project to develop repopack.com itself, we could build in `rp report` as a shortcut to `rp task create --project repopack`
  58. blue btw, that EDITOR thing, I get it, but it's probably not gonna work with AI. maybe `rp task create --title <title> [--editor | --stdin | --text <text> | --file <path>]
  59. dreamreal I think it would: --no-editor and --editor, the LLM would use --no-editor if it has no body content
  60. blue I believe it's currently possible to create tasks without content, maybe that should be rejected
  61. dreamreal I certainly did it :D
  62. dreamreal was surprising, which is why I filed the issue
  63. * phaleth quit (Quit: phaleth)
  64. blue will amend to enforce
  65. blue only title makes no sense
  66. blue as for --no-editor; I'm generally not a *huge* fan of negative options
  67. blue we can offer --editor, --stdin, --text and --file and enforce EXACTLY one
  68. blue doing --editor as a default will trip up llm hard
  69. blue it's better to do no default and force verbosity here
  70. blue don't you agree?
  71. dreamreal Sure
  72. blue k
  73. blue sorry dreamreal, I didn't have the flog project subscribed, so didn't see
  74. blue it's now deployed
  75. blue we have individual changelog pages for primate packages now, e.g. https://primate.run/api/changelog/pema
  76. dreamreal nice!
  77. blue lioloc: fyi https://primate.run/blog/primate-045#acknowledgements
  78. nevet Primate — Primate 0.45: SQL transactions, Kotlin backend and iOS apps
  79. blue dreamreal: I have a question...
  80. blue I think this would make nevet cooler, but maybe a bother to implement
  81. blue do you think it should follow anchor tags and show their content?
  82. blue the problem nevet would face is that the id-bearing element might content a bit more than needed
  83. blue for example:
  84. blue on the url I just posted, `document.querySelector("#acknowledgements").textContent` yields
  85. blue '\n Acknowledgements\n \n \n \n \n \n \n \n '
  86. blue document.querySelector("#acknowledgements").textContent.trim();
  87. blue 'Acknowledgements'
  88. blue that could work.
  89. blue so you could have:
  90. blue Primate — Primate 0.45: SQL transactions, Kotlin backend and iOS apps, Acknowledgements
  91. blue or () or whatever separator of your using
  92. blue thoughts? should I create task?
  93. blue might contain*
  94. dreamreal Hmm
  95. dreamreal following anchor tags might be a drag
  96. dreamreal Will think about it though. The problem is that anchors don't lend themselves to *titles*
  97. blue k
  98. blue dreamreal: repopack will switch to the following session model: web sessions run for 7d, indefinitely extensible (by action). rp sessions run for 30d, same. In other words: you do nothing for 7d on the web, you get logged out. Same for rp, but 30d.
  99. blue please ack if you consider this model productive, otherwise counteroffer
  100. blue the reason the web ui is more limited is that while offering less actions, it's more powerful: all guarded actions run through the web ui, and they currently do NOT escalate to 2nd factor. In the future, when we can escalate to 2nd factor on guarded actions, those web sessions might be extended.

Search posts